home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: vmsnet.vms-posix
- Path: sparky!uunet!paladin.american.edu!europa.asd.contel.com!darwin.sura.net!zaphod.mps.ohio-state.edu!uwm.edu!src.honeywell.com!iasdvc.honeywell.com!kish
- From: kish@iasdvc.honeywell.com ()
- Subject: Re: name of VMS product has changed...
- Message-ID: <1992Aug18.223322.25145@src.honeywell.com>
- Lines: 52
- Sender: news@src.honeywell.com (News interface)
- Nntp-Posting-Host: iasdv1.iasd.honeywell.com
- Reply-To: kish@iasdvc.honeywell.com ()
- Organization: HONEYWELL_IAC
- References: <2415@mitech.com> <1992Aug18.091028@mccall.com>
- Date: Tue, 18 Aug 1992 22:33:22 GMT
-
-
- >[Followups to vmsnet.vms-posix]
- |>
- |>They could start by implementing the things that they said in a decus session
- |>that they were going to implement. They went to a lot of trouble to set up a
- |>satellite link to Italy so the developer could tell us what was in VMS Posix.
- |>One of them was unix-style light-weight processes (complete with a cute drawing
- |>likening a unix process to a sparrow and a VMS process to a vulture). I can
- |>understand them wanting to get things out the door, but DEC hasn't acknowleged
- |>anywhere I've seen that the performance stinks, and they haven't said they are
- |>going to do anything about it. They are talking about Posix 1.1 as if it were
- |>out, and it isn't, and they still haven't said whether it fixes any performance
- |>problems (obviously, since they haven't acknowleged any performance problems).
- |>
- I saw a press release which has 1.1 in release the end of this month
- ( Insert your own humous comment here).....
-
- I did do some checking on a few of the performance issues....
- "Compute bound process run slow"
- I ran a c version of the whestone benchmark and it ran at a apparent rate
- of .5 of the VMS copy. As it turns out it was the fact that POSIX
- does dynamic links to the RTL. In this case I was calling a bunch
- of functions which each had to be linked on call ( at least the first time).
- This distorted the numbers to give the idea that compute bound process
- run slow. Actally it turns out the intuitive answer is correct they run
- at the process speed.
-
- I did some similar testing for "FORKED" things which did not look as bad
- as native VMS immages but still no where close to ULTRIX. Which to be honest
- was really all I expected.
-
- |>Has anyone else noticed that lately all the marketting hype and even published
- |>benchmarks are for software versions that aren't out yet? I've been reading
- |>about Rdb 4.1 for a long time, I think since last year. It isn't out yet either,
- |>but DEC has already published TPS numbers to convince everyone how great it is.
- |>Slimy marketting.
- |>--
-
- We have 4.1 check your mail box! :) It just hit our door.
-
- I dont know about |>Slimy marketting or just a long process between field test
- and shipping. We do nearly as bad here shipping VAX software so perhaps
- I am more aligned to the shipping side. Still I agree we could all do better.
- Remeber DECSET for VMS announced for shipping Middle of LAST year and still
- not at the gate! AGHHHHHHHHHHH
-